Other sources of bias might include severe baseline imbalances, blocked randomization in unblinded trials (can lead to prediction of future assignments), or study design characteristics. Possible design-specific biases to look out for include having an inadequate wash-out period in a cross-over trial, or unit of analysis errors in cluster trials.
Tools
Guidance for minimizing bias in cluster randomized trials and cross-over trials: Cochrane Handbook, Chapter 16: Special Topics in Statistics
Analysis of cluster randomized trials: One major difference in the analysis of a cluster randomized trial compared to a parallel randomized controlled trial is that the unit of analysis is no longer the individual participant, it is the group to which the participant belongs. This group, or cluster, could be something like a hospital, school, or community. Since the group is what has been randomized, the group is what must be analyzed. A simple summary of these concepts can be found here. Another consideration is that members of a group are likely more similar to each other than to members of other groups, therefore this correlation needs to be accounted for using the intracluster correlation coefficient (ICC).
Analysis of cross-over trials: Since each participant in a cross-over trial will have two measurements, one for the effect of the intervention and one for the effect of the control, the appropriate analysis is a paired t-test.
On this page we've compiled a number of examples of risk of bias assessments - the good, the bad, and those that are a bit unclear. Feel free to work through them yourself and come up with an assessment oflow,unclear, orhighrisk of bias (our judgments and rationale are on theassessments page), or download a spreadsheet file with the same information. RoB assessments are divided up into the seven major domains: sequence generation, allocation concelment, blinding of participants/personnel, blinding of outcome assessors, incomplete outcome data, selective outcome reporting, and other sources of bias. A quotation is given with the article title following in brackets.
If you have other examples, please add them to the list!
*AOM: acute otitis media; RAD: reactive airway disease; full citation: Continuous twice daily or once daily amoxicillin prophylaxis compared with placebo for children with recurrent acute otitis media
Lexchin J, Bero LA, Djulbegovic B, et al. Pharmaceutical industry sponsorship and research outcome and quality: systematic review. BMJ 2003;326:1167-1170. [PubMed]
Bekelman JE, Li Y, Gross CP. Scope and impact of financial conflicts of interest in biomedical research: a systematic review. JAMA 2003;298:454-465. [PubMed]
Montori VM, Devereaux PJ, Adhikari NK, et al. Randomized trials stopped early for benefit: a systematic review. JAMA 2005;294:2203-2209. [PubMed]
Bassler D, Ferreira-Gonzalez I, Briel M, et al. Systematic reviewers neglect bias that results from trials stopped early for benefit. J Clin Epidemiol 2007;60:869-873. [PubMed]
Sismondo S. Pharmaceutical company funding and its consequences: a qualitative systematic review. Contemp Clin Trials 2008;29:109-113. [PubMed]
What are other sources of bias?
Table of Contents
Tools
Analysis of cluster randomized trials: One major difference in the analysis of a cluster randomized trial compared to a parallel randomized controlled trial is that the unit of analysis is no longer the individual participant, it is the group to which the participant belongs. This group, or cluster, could be something like a hospital, school, or community. Since the group is what has been randomized, the group is what must be analyzed. A simple summary of these concepts can be found here. Another consideration is that members of a group are likely more similar to each other than to members of other groups, therefore this correlation needs to be accounted for using the intracluster correlation coefficient (ICC).
Analysis of cross-over trials: Since each participant in a cross-over trial will have two measurements, one for the effect of the intervention and one for the effect of the control, the appropriate analysis is a paired t-test.
[back to top]
Examples
On this page we've compiled a number of examples of risk of bias assessments - the good, the bad, and those that are a bit unclear. Feel free to work through them yourself and come up with an assessment of low, unclear, or high risk of bias (our judgments and rationale are on the assessments page), or download a spreadsheet file with the same information. RoB assessments are divided up into the seven major domains: sequence generation, allocation concelment, blinding of participants/personnel, blinding of outcome assessors, incomplete outcome data, selective outcome reporting, and other sources of bias. A quotation is given with the article title following in brackets.
If you have other examples, please add them to the list!
Risk of Bias Guidelines
Download examples:
*AOM: acute otitis media; RAD: reactive airway disease; full citation: Continuous twice daily or once daily amoxicillin prophylaxis compared with placebo for children with recurrent acute otitis media
[back to top] [RoB Assessment Page]
References
[back to top]